2 * Copyright (C) International Business Machines Corp., 2000-2004
3 * Portions Copyright (C) Christoph Hellwig, 2001-2002
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
13 * the GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 #include <linux/module.h>
22 #include <linux/parser.h>
23 #include <linux/completion.h>
24 #include <linux/vfs.h>
25 #include <linux/quotaops.h>
26 #include <linux/mount.h>
27 #include <linux/moduleparam.h>
28 #include <linux/kthread.h>
29 #include <linux/posix_acl.h>
30 #include <linux/buffer_head.h>
31 #include <linux/exportfs.h>
32 #include <linux/crc32.h>
33 #include <linux/slab.h>
34 #include <asm/uaccess.h>
35 #include <linux/seq_file.h>
37 #include "jfs_incore.h"
38 #include "jfs_filsys.h"
39 #include "jfs_inode.h"
40 #include "jfs_metapage.h"
41 #include "jfs_superblock.h"
45 #include "jfs_debug.h"
47 MODULE_DESCRIPTION("The Journaled Filesystem (JFS)");
48 MODULE_AUTHOR("Steve Best/Dave Kleikamp/Barry Arndt, IBM");
49 MODULE_LICENSE("GPL");
51 static struct kmem_cache * jfs_inode_cachep;
53 static const struct super_operations jfs_super_operations;
54 static const struct export_operations jfs_export_operations;
55 static struct file_system_type jfs_fs_type;
57 #define MAX_COMMIT_THREADS 64
58 static int commit_threads = 0;
59 module_param(commit_threads, int, 0);
60 MODULE_PARM_DESC(commit_threads, "Number of commit threads");
62 static struct task_struct *jfsCommitThread[MAX_COMMIT_THREADS];
63 struct task_struct *jfsIOthread;
64 struct task_struct *jfsSyncThread;
66 #ifdef CONFIG_JFS_DEBUG
67 int jfsloglevel = JFS_LOGLEVEL_WARN;
68 module_param(jfsloglevel, int, 0644);
69 MODULE_PARM_DESC(jfsloglevel, "Specify JFS loglevel (0, 1 or 2)");
72 static void jfs_handle_error(struct super_block *sb)
74 struct jfs_sb_info *sbi = JFS_SBI(sb);
76 if (sb->s_flags & MS_RDONLY)
79 updateSuper(sb, FM_DIRTY);
81 if (sbi->flag & JFS_ERR_PANIC)
82 panic("JFS (device %s): panic forced after error\n",
84 else if (sbi->flag & JFS_ERR_REMOUNT_RO) {
85 jfs_err("ERROR: (device %s): remounting filesystem "
88 sb->s_flags |= MS_RDONLY;
91 /* nothing is done for continue beyond marking the superblock dirty */
94 void jfs_error(struct super_block *sb, const char * function, ...)
96 static char error_buf[256];
99 va_start(args, function);
100 vsnprintf(error_buf, sizeof(error_buf), function, args);
103 printk(KERN_ERR "ERROR: (device %s): %s\n", sb->s_id, error_buf);
105 jfs_handle_error(sb);
108 static struct inode *jfs_alloc_inode(struct super_block *sb)
110 struct jfs_inode_info *jfs_inode;
112 jfs_inode = kmem_cache_alloc(jfs_inode_cachep, GFP_NOFS);
115 return &jfs_inode->vfs_inode;
118 static void jfs_i_callback(struct rcu_head *head)
120 struct inode *inode = container_of(head, struct inode, i_rcu);
121 struct jfs_inode_info *ji = JFS_IP(inode);
122 kmem_cache_free(jfs_inode_cachep, ji);
125 static void jfs_destroy_inode(struct inode *inode)
127 struct jfs_inode_info *ji = JFS_IP(inode);
129 BUG_ON(!list_empty(&ji->anon_inode_list));
131 spin_lock_irq(&ji->ag_lock);
132 if (ji->active_ag != -1) {
133 struct bmap *bmap = JFS_SBI(inode->i_sb)->bmap;
134 atomic_dec(&bmap->db_active[ji->active_ag]);
137 spin_unlock_irq(&ji->ag_lock);
138 call_rcu(&inode->i_rcu, jfs_i_callback);
141 static int jfs_statfs(struct dentry *dentry, struct kstatfs *buf)
143 struct jfs_sb_info *sbi = JFS_SBI(dentry->d_sb);
145 struct inomap *imap = JFS_IP(sbi->ipimap)->i_imap;
147 jfs_info("In jfs_statfs");
148 buf->f_type = JFS_SUPER_MAGIC;
149 buf->f_bsize = sbi->bsize;
150 buf->f_blocks = sbi->bmap->db_mapsize;
151 buf->f_bfree = sbi->bmap->db_nfree;
152 buf->f_bavail = sbi->bmap->db_nfree;
154 * If we really return the number of allocated & free inodes, some
155 * applications will fail because they won't see enough free inodes.
156 * We'll try to calculate some guess as to how may inodes we can
159 * buf->f_files = atomic_read(&imap->im_numinos);
160 * buf->f_ffree = atomic_read(&imap->im_numfree);
162 maxinodes = min((s64) atomic_read(&imap->im_numinos) +
163 ((sbi->bmap->db_nfree >> imap->im_l2nbperiext)
164 << L2INOSPEREXT), (s64) 0xffffffffLL);
165 buf->f_files = maxinodes;
166 buf->f_ffree = maxinodes - (atomic_read(&imap->im_numinos) -
167 atomic_read(&imap->im_numfree));
168 buf->f_fsid.val[0] = (u32)crc32_le(0, sbi->uuid, sizeof(sbi->uuid)/2);
169 buf->f_fsid.val[1] = (u32)crc32_le(0, sbi->uuid + sizeof(sbi->uuid)/2,
170 sizeof(sbi->uuid)/2);
172 buf->f_namelen = JFS_NAME_MAX;
176 static void jfs_put_super(struct super_block *sb)
178 struct jfs_sb_info *sbi = JFS_SBI(sb);
181 jfs_info("In jfs_put_super");
183 dquot_disable(sb, -1, DQUOT_USAGE_ENABLED | DQUOT_LIMITS_ENABLED);
187 jfs_err("jfs_umount failed with return code %d", rc);
189 unload_nls(sbi->nls_tab);
191 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
192 iput(sbi->direct_inode);
198 Opt_integrity, Opt_nointegrity, Opt_iocharset, Opt_resize,
199 Opt_resize_nosize, Opt_errors, Opt_ignore, Opt_err, Opt_quota,
200 Opt_usrquota, Opt_grpquota, Opt_uid, Opt_gid, Opt_umask
203 static const match_table_t tokens = {
204 {Opt_integrity, "integrity"},
205 {Opt_nointegrity, "nointegrity"},
206 {Opt_iocharset, "iocharset=%s"},
207 {Opt_resize, "resize=%u"},
208 {Opt_resize_nosize, "resize"},
209 {Opt_errors, "errors=%s"},
210 {Opt_ignore, "noquota"},
211 {Opt_ignore, "quota"},
212 {Opt_usrquota, "usrquota"},
213 {Opt_grpquota, "grpquota"},
216 {Opt_umask, "umask=%u"},
220 static int parse_options(char *options, struct super_block *sb, s64 *newLVSize,
223 void *nls_map = (void *)-1; /* -1: no change; NULL: none */
225 struct jfs_sb_info *sbi = JFS_SBI(sb);
232 while ((p = strsep(&options, ",")) != NULL) {
233 substring_t args[MAX_OPT_ARGS];
238 token = match_token(p, tokens, args);
241 *flag &= ~JFS_NOINTEGRITY;
243 case Opt_nointegrity:
244 *flag |= JFS_NOINTEGRITY;
247 /* Silently ignore the quota options */
248 /* Don't do anything ;-) */
251 if (nls_map && nls_map != (void *) -1)
253 if (!strcmp(args[0].from, "none"))
256 nls_map = load_nls(args[0].from);
259 "JFS: charset not found\n");
266 char *resize = args[0].from;
267 *newLVSize = simple_strtoull(resize, &resize, 0);
270 case Opt_resize_nosize:
272 *newLVSize = sb->s_bdev->bd_inode->i_size >>
273 sb->s_blocksize_bits;
276 "JFS: Cannot determine volume size\n");
281 char *errors = args[0].from;
282 if (!errors || !*errors)
284 if (!strcmp(errors, "continue")) {
285 *flag &= ~JFS_ERR_REMOUNT_RO;
286 *flag &= ~JFS_ERR_PANIC;
287 *flag |= JFS_ERR_CONTINUE;
288 } else if (!strcmp(errors, "remount-ro")) {
289 *flag &= ~JFS_ERR_CONTINUE;
290 *flag &= ~JFS_ERR_PANIC;
291 *flag |= JFS_ERR_REMOUNT_RO;
292 } else if (!strcmp(errors, "panic")) {
293 *flag &= ~JFS_ERR_CONTINUE;
294 *flag &= ~JFS_ERR_REMOUNT_RO;
295 *flag |= JFS_ERR_PANIC;
298 "JFS: %s is an invalid error handler\n",
308 *flag |= JFS_USRQUOTA;
311 *flag |= JFS_GRPQUOTA;
318 "JFS: quota operations not supported\n");
323 char *uid = args[0].from;
324 sbi->uid = simple_strtoul(uid, &uid, 0);
329 char *gid = args[0].from;
330 sbi->gid = simple_strtoul(gid, &gid, 0);
335 char *umask = args[0].from;
336 sbi->umask = simple_strtoul(umask, &umask, 8);
337 if (sbi->umask & ~0777) {
339 "JFS: Invalid value of umask\n");
345 printk("jfs: Unrecognized mount option \"%s\" "
346 " or missing value\n", p);
351 if (nls_map != (void *) -1) {
352 /* Discard old (if remount) */
353 unload_nls(sbi->nls_tab);
354 sbi->nls_tab = nls_map;
359 if (nls_map && nls_map != (void *) -1)
364 static int jfs_remount(struct super_block *sb, int *flags, char *data)
368 int flag = JFS_SBI(sb)->flag;
371 if (!parse_options(data, sb, &newLVSize, &flag)) {
376 if (sb->s_flags & MS_RDONLY) {
378 "JFS: resize requires volume to be mounted read-write\n");
381 rc = jfs_extendfs(sb, newLVSize, 0);
386 if ((sb->s_flags & MS_RDONLY) && !(*flags & MS_RDONLY)) {
388 * Invalidate any previously read metadata. fsck may have
389 * changed the on-disk data since we mounted r/o
391 truncate_inode_pages(JFS_SBI(sb)->direct_inode->i_mapping, 0);
393 JFS_SBI(sb)->flag = flag;
394 ret = jfs_mount_rw(sb, 1);
396 /* mark the fs r/w for quota activity */
397 sb->s_flags &= ~MS_RDONLY;
399 dquot_resume(sb, -1);
402 if ((!(sb->s_flags & MS_RDONLY)) && (*flags & MS_RDONLY)) {
403 rc = dquot_suspend(sb, -1);
407 rc = jfs_umount_rw(sb);
408 JFS_SBI(sb)->flag = flag;
411 if ((JFS_SBI(sb)->flag & JFS_NOINTEGRITY) != (flag & JFS_NOINTEGRITY))
412 if (!(sb->s_flags & MS_RDONLY)) {
413 rc = jfs_umount_rw(sb);
417 JFS_SBI(sb)->flag = flag;
418 ret = jfs_mount_rw(sb, 1);
421 JFS_SBI(sb)->flag = flag;
426 static int jfs_fill_super(struct super_block *sb, void *data, int silent)
428 struct jfs_sb_info *sbi;
432 int flag, ret = -EINVAL;
434 jfs_info("In jfs_read_super: s_flags=0x%lx", sb->s_flags);
436 if (!new_valid_dev(sb->s_bdev->bd_dev))
439 sbi = kzalloc(sizeof (struct jfs_sb_info), GFP_KERNEL);
444 sb->s_max_links = JFS_LINK_MAX;
446 sbi->uid = sbi->gid = sbi->umask = -1;
448 /* initialize the mount flag and determine the default error handler */
449 flag = JFS_ERR_REMOUNT_RO;
451 if (!parse_options((char *) data, sb, &newLVSize, &flag))
455 #ifdef CONFIG_JFS_POSIX_ACL
456 sb->s_flags |= MS_POSIXACL;
460 printk(KERN_ERR "resize option for remount only\n");
465 * Initialize blocksize to 4K.
467 sb_set_blocksize(sb, PSIZE);
470 * Set method vectors.
472 sb->s_op = &jfs_super_operations;
473 sb->s_export_op = &jfs_export_operations;
475 sb->dq_op = &dquot_operations;
476 sb->s_qcop = &dquot_quotactl_ops;
480 * Initialize direct-mapping inode/address-space
482 inode = new_inode(sb);
488 inode->i_size = sb->s_bdev->bd_inode->i_size;
489 inode->i_mapping->a_ops = &jfs_metapage_aops;
490 insert_inode_hash(inode);
491 mapping_set_gfp_mask(inode->i_mapping, GFP_NOFS);
493 sbi->direct_inode = inode;
498 jfs_err("jfs_mount failed w/return code = %d", rc);
500 goto out_mount_failed;
502 if (sb->s_flags & MS_RDONLY)
505 rc = jfs_mount_rw(sb, 0);
508 jfs_err("jfs_mount_rw failed, return code = %d",
515 sb->s_magic = JFS_SUPER_MAGIC;
517 if (sbi->mntflag & JFS_OS2)
518 sb->s_d_op = &jfs_ci_dentry_operations;
520 inode = jfs_iget(sb, ROOT_I);
522 ret = PTR_ERR(inode);
525 sb->s_root = d_make_root(inode);
529 /* logical blocks are represented by 40 bits in pxd_t, etc. */
530 sb->s_maxbytes = ((u64) sb->s_blocksize) << 40;
531 #if BITS_PER_LONG == 32
533 * Page cache is indexed by long.
534 * I would use MAX_LFS_FILESIZE, but it's only half as big
536 sb->s_maxbytes = min(((u64) PAGE_CACHE_SIZE << 32) - 1, (u64)sb->s_maxbytes);
542 jfs_err("jfs_read_super: get root dentry failed");
547 jfs_err("jfs_umount failed with return code %d", rc);
550 filemap_write_and_wait(sbi->direct_inode->i_mapping);
551 truncate_inode_pages(sbi->direct_inode->i_mapping, 0);
552 make_bad_inode(sbi->direct_inode);
553 iput(sbi->direct_inode);
554 sbi->direct_inode = NULL;
557 unload_nls(sbi->nls_tab);
563 static int jfs_freeze(struct super_block *sb)
565 struct jfs_sb_info *sbi = JFS_SBI(sb);
566 struct jfs_log *log = sbi->log;
568 if (!(sb->s_flags & MS_RDONLY)) {
571 updateSuper(sb, FM_CLEAN);
576 static int jfs_unfreeze(struct super_block *sb)
578 struct jfs_sb_info *sbi = JFS_SBI(sb);
579 struct jfs_log *log = sbi->log;
582 if (!(sb->s_flags & MS_RDONLY)) {
583 updateSuper(sb, FM_MOUNT);
584 if ((rc = lmLogInit(log)))
585 jfs_err("jfs_unlock failed with return code %d", rc);
592 static struct dentry *jfs_do_mount(struct file_system_type *fs_type,
593 int flags, const char *dev_name, void *data)
595 return mount_bdev(fs_type, flags, dev_name, data, jfs_fill_super);
598 static int jfs_sync_fs(struct super_block *sb, int wait)
600 struct jfs_log *log = JFS_SBI(sb)->log;
602 /* log == NULL indicates read-only mount */
604 jfs_flush_journal(log, wait);
611 static int jfs_show_options(struct seq_file *seq, struct dentry *root)
613 struct jfs_sb_info *sbi = JFS_SBI(root->d_sb);
616 seq_printf(seq, ",uid=%d", sbi->uid);
618 seq_printf(seq, ",gid=%d", sbi->gid);
619 if (sbi->umask != -1)
620 seq_printf(seq, ",umask=%03o", sbi->umask);
621 if (sbi->flag & JFS_NOINTEGRITY)
622 seq_puts(seq, ",nointegrity");
624 seq_printf(seq, ",iocharset=%s", sbi->nls_tab->charset);
625 if (sbi->flag & JFS_ERR_CONTINUE)
626 seq_printf(seq, ",errors=continue");
627 if (sbi->flag & JFS_ERR_PANIC)
628 seq_printf(seq, ",errors=panic");
631 if (sbi->flag & JFS_USRQUOTA)
632 seq_puts(seq, ",usrquota");
634 if (sbi->flag & JFS_GRPQUOTA)
635 seq_puts(seq, ",grpquota");
643 /* Read data from quotafile - avoid pagecache and such because we cannot afford
644 * acquiring the locks... As quota files are never truncated and quota code
645 * itself serializes the operations (and no one else should touch the files)
646 * we don't have to be afraid of races */
647 static ssize_t jfs_quota_read(struct super_block *sb, int type, char *data,
648 size_t len, loff_t off)
650 struct inode *inode = sb_dqopt(sb)->files[type];
651 sector_t blk = off >> sb->s_blocksize_bits;
653 int offset = off & (sb->s_blocksize - 1);
656 struct buffer_head tmp_bh;
657 struct buffer_head *bh;
658 loff_t i_size = i_size_read(inode);
662 if (off+len > i_size)
666 tocopy = sb->s_blocksize - offset < toread ?
667 sb->s_blocksize - offset : toread;
670 tmp_bh.b_size = 1 << inode->i_blkbits;
671 err = jfs_get_block(inode, blk, &tmp_bh, 0);
674 if (!buffer_mapped(&tmp_bh)) /* A hole? */
675 memset(data, 0, tocopy);
677 bh = sb_bread(sb, tmp_bh.b_blocknr);
680 memcpy(data, bh->b_data+offset, tocopy);
691 /* Write to quotafile */
692 static ssize_t jfs_quota_write(struct super_block *sb, int type,
693 const char *data, size_t len, loff_t off)
695 struct inode *inode = sb_dqopt(sb)->files[type];
696 sector_t blk = off >> sb->s_blocksize_bits;
698 int offset = off & (sb->s_blocksize - 1);
700 size_t towrite = len;
701 struct buffer_head tmp_bh;
702 struct buffer_head *bh;
704 mutex_lock(&inode->i_mutex);
705 while (towrite > 0) {
706 tocopy = sb->s_blocksize - offset < towrite ?
707 sb->s_blocksize - offset : towrite;
710 tmp_bh.b_size = 1 << inode->i_blkbits;
711 err = jfs_get_block(inode, blk, &tmp_bh, 1);
714 if (offset || tocopy != sb->s_blocksize)
715 bh = sb_bread(sb, tmp_bh.b_blocknr);
717 bh = sb_getblk(sb, tmp_bh.b_blocknr);
723 memcpy(bh->b_data+offset, data, tocopy);
724 flush_dcache_page(bh->b_page);
725 set_buffer_uptodate(bh);
726 mark_buffer_dirty(bh);
735 if (len == towrite) {
736 mutex_unlock(&inode->i_mutex);
739 if (inode->i_size < off+len-towrite)
740 i_size_write(inode, off+len-towrite);
742 inode->i_mtime = inode->i_ctime = CURRENT_TIME;
743 mark_inode_dirty(inode);
744 mutex_unlock(&inode->i_mutex);
745 return len - towrite;
750 static const struct super_operations jfs_super_operations = {
751 .alloc_inode = jfs_alloc_inode,
752 .destroy_inode = jfs_destroy_inode,
753 .dirty_inode = jfs_dirty_inode,
754 .write_inode = jfs_write_inode,
755 .evict_inode = jfs_evict_inode,
756 .put_super = jfs_put_super,
757 .sync_fs = jfs_sync_fs,
758 .freeze_fs = jfs_freeze,
759 .unfreeze_fs = jfs_unfreeze,
760 .statfs = jfs_statfs,
761 .remount_fs = jfs_remount,
762 .show_options = jfs_show_options,
764 .quota_read = jfs_quota_read,
765 .quota_write = jfs_quota_write,
769 static const struct export_operations jfs_export_operations = {
770 .fh_to_dentry = jfs_fh_to_dentry,
771 .fh_to_parent = jfs_fh_to_parent,
772 .get_parent = jfs_get_parent,
775 static struct file_system_type jfs_fs_type = {
776 .owner = THIS_MODULE,
778 .mount = jfs_do_mount,
779 .kill_sb = kill_block_super,
780 .fs_flags = FS_REQUIRES_DEV,
783 static void init_once(void *foo)
785 struct jfs_inode_info *jfs_ip = (struct jfs_inode_info *) foo;
787 memset(jfs_ip, 0, sizeof(struct jfs_inode_info));
788 INIT_LIST_HEAD(&jfs_ip->anon_inode_list);
789 init_rwsem(&jfs_ip->rdwrlock);
790 mutex_init(&jfs_ip->commit_mutex);
791 init_rwsem(&jfs_ip->xattr_sem);
792 spin_lock_init(&jfs_ip->ag_lock);
793 jfs_ip->active_ag = -1;
794 inode_init_once(&jfs_ip->vfs_inode);
797 static int __init init_jfs_fs(void)
803 kmem_cache_create("jfs_ip", sizeof(struct jfs_inode_info), 0,
804 SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD,
806 if (jfs_inode_cachep == NULL)
810 * Metapage initialization
812 rc = metapage_init();
814 jfs_err("metapage_init failed w/rc = %d", rc);
819 * Transaction Manager initialization
823 jfs_err("txInit failed w/rc = %d", rc);
828 * I/O completion thread (endio)
830 jfsIOthread = kthread_run(jfsIOWait, NULL, "jfsIO");
831 if (IS_ERR(jfsIOthread)) {
832 rc = PTR_ERR(jfsIOthread);
833 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
837 if (commit_threads < 1)
838 commit_threads = num_online_cpus();
839 if (commit_threads > MAX_COMMIT_THREADS)
840 commit_threads = MAX_COMMIT_THREADS;
842 for (i = 0; i < commit_threads; i++) {
843 jfsCommitThread[i] = kthread_run(jfs_lazycommit, NULL, "jfsCommit");
844 if (IS_ERR(jfsCommitThread[i])) {
845 rc = PTR_ERR(jfsCommitThread[i]);
846 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
848 goto kill_committask;
852 jfsSyncThread = kthread_run(jfs_sync, NULL, "jfsSync");
853 if (IS_ERR(jfsSyncThread)) {
854 rc = PTR_ERR(jfsSyncThread);
855 jfs_err("init_jfs_fs: fork failed w/rc = %d", rc);
856 goto kill_committask;
863 rc = register_filesystem(&jfs_fs_type);
870 kthread_stop(jfsSyncThread);
872 for (i = 0; i < commit_threads; i++)
873 kthread_stop(jfsCommitThread[i]);
874 kthread_stop(jfsIOthread);
880 kmem_cache_destroy(jfs_inode_cachep);
884 static void __exit exit_jfs_fs(void)
888 jfs_info("exit_jfs_fs called");
893 kthread_stop(jfsIOthread);
894 for (i = 0; i < commit_threads; i++)
895 kthread_stop(jfsCommitThread[i]);
896 kthread_stop(jfsSyncThread);
900 unregister_filesystem(&jfs_fs_type);
901 kmem_cache_destroy(jfs_inode_cachep);
904 module_init(init_jfs_fs)
905 module_exit(exit_jfs_fs)